There's nothing special about pulse, right?  This is just "the anchor node may animate" (which is in turn a special case of "the webpage may move the anchor node deliberately).

Yes, there's nothing special about pulse. But, I think this problem exists regardless of whether the anchor node animates. We could anchor to something below the pulse node, but the pulse may cause a change in the position of the anchor and unnecessary anchoring.

Its a special case of "we don't know whether the layout is caused by something loading, or an intentional change". It may actually not be that big of an issue in practice because these animations are services on cc. I'm deliberately changing 'top' to trigger a layout. 

But I filed this to be aware of the types of problems that might happen. I think the right fix here would be to opt-out the element containing the pulsing eleming.

I think this was fixed by SANACLAP, since "transform" is a suppression trigger.